Alcohol is often implicated in automobile deaths. According to police
reports, at least one driver has been drinking (although not necessarily
over the legal blood-alcohol limit) in over 30 percent of fatal crashes.
During the time periods in which alcohol usage is greatest, that proportion
rises to almost 60 percent.1
Without knowing the fraction of drivers on the road who have been
drinking, however, one cannot possibly draw conclusions about the relative
fatal crash risk of drinking versus sober drivers, the externality
associated with drinking and driving, or the appropriate public policy
response. For instance, if 30 percent of the drivers had been drinking,
over half of all two-vehicle crashes would be expected to involve at least
one drinking driver, even if drinking drivers were no more dangerous
than sober drivers.